Electric power can be produced by a variety of types of generating units, with differences in costs of preparing units to produce, time lags to prepare units to produce, and costs of production. Because of cost differences among units, decisions made will affect overall costs incurred. Because of time lags, decisions are based on electricity demand forecasts, which in turn depend importantly on weather forecasts.

This paper presents estimates of the value of weather forecasts in making decisions to commit electric generating units for production. Four weather forecasts are considered: (1) a Persistence forecast (no change in weather), (2) a computer generated National Weather Service (NWS) forecast known as MAV, (3) the “official” NWS forecast reflecting typical forecaster improvements over the MAV, and (4) a Perfect forecast (possible since this is a retrospective study). For each of these weather forecasts, electricity demand forecasts are produced using a model like that used by electric utilities. Then, for each demand forecast, production costs are estimated. This is done for each of six sites around the US, and the results aggregated.

Based on the production costs differences estimated, the overall annual cost saving generated by official NWS forecasts is $155 million. The potential additional cost saving obtainable from a Perfect forecast is $69 million per year. It is estimated that an additional 1 percent improvement in forecast quality (from the official NWS forecast) would be worth an additional $1.3 million per year.

This article presents information about the economic value of services provided by the US National Weather Service (NWS). Such services are provided to the public at no charge and paid for by the US government out of available resources. Since there are many competing demands on the government resources, it is important to know the values generated by current or prospective uses of these resources. This knowledge makes it possible to achieve the greatest overall benefits from public resources. Since many BAMS readers will have a special interest in activities of the NWS, the value of NWS services would be of particular interest to them.

It is expected that the article will be approximately 4000 words in length. There would be one figure, in black and white. This figure would be generated using an Excel spreadsheet. No use of electronic supplements is anticipated.